There is an inconsistency in the quadprog wrapper when unwrapping dual multipliers:
|
if (meq>0 and n_in==0): |
|
y_sol = -y |
|
if (meq>0 and n_in>0): |
|
y_sol[eq_ids] = y[:meq] |
That is, y_sol[eq_ids] is taken as -y when there is no inequality but +y when there are some.
I think +y is the correct conversion in this context, but to be checked.
There is an inconsistency in the quadprog wrapper when unwrapping dual multipliers:
proxqp_benchmark/solvers/quadprog.py
Lines 86 to 89 in 240c3f3
That is,
y_sol[eq_ids]is taken as-ywhen there is no inequality but+ywhen there are some.I think
+yis the correct conversion in this context, but to be checked.